The emergence of generative artificial intelligence (GenAI) in education has revived the debate over control, autonomy, and equity. A recent scoping review focusing on the relationship between GenAI and agency –the ability of students and teachers to make meaningful decisions in their learning process– reveals complex findings. While personalization offered by virtual assistants can empower learners, there is also a risk of deepening existing gaps and reducing autonomy if robust ethical and technical frameworks are not implemented. In this context, edtech companies and software developers have a responsibility to design solutions that balance transformative potential with the protection of human agency.
From a technical perspective, integrating GenAI into educational platforms requires a careful approach to custom software development that respects user-centered design principles. It is not simply about adding a chatbot; an architecture is needed that allows teachers to configure the level of AI intervention, while the student maintains control over their learning path. For example, AI agents can offer personalized recommendations based on performance, but always with the option for the learner to ignore or modify them. This decision-making ability is key to preserving agency.
The review identifies three major themes: control in digital spaces, variable engagement and access, and changing notions of agency. In the first theme, control manifests not only in who decides what content is shown, but also in data governance. Platforms using AWS/Azure cloud, for instance, must ensure that student data is not commercially exploited and that algorithms are auditable. Here cybersecurity plays a crucial role: protecting sensitive information and avoiding algorithmic bias requires implementing measures such as end-to-end encryption and data anonymization. A company like Q2BSTUDIO, specialized in security, can advise on best practices to secure these environments.
The second theme, variable engagement and access, highlights the digital divide. While some students benefit from instant help from a virtual tutor, others lack connectivity or adequate devices. BI/Power BI solutions can help institutions visualize these inequalities and design targeted interventions. For example, a real-time dashboard can show which students interact less with the platform, allowing teachers to offer additional support. Moreover, process automation tools can free up teacher time for personalized attention, provided the technological design does not impose additional barriers.
The third theme, changing notions of agency, addresses how AI redefines the role of teacher and student. Traditionally, teacher agency involved designing curriculum and assessment; now, with systems that automatically generate exercises and feedback, the teacher must learn to curate and supervise those outputs. Here, well-designed AI agents can act as assistants, not substitutes. For instance, an AI system that suggests exam questions can be reviewed and adjusted by the teacher, maintaining pedagogical authority. To achieve this, the platform must offer intuitive interfaces and customization tools that allow the teacher to modify AI parameters without deep technical knowledge.
From a business perspective, developing educational software with GenAI must prioritize transparency. Algorithms should not be black boxes; users –both teachers and students– need to understand why a resource or activity is recommended. Explainable AI (XAI) techniques are essential. Furthermore, integration with cloud services such as AWS or Azure allows scaling solutions and ensuring availability, but also requires careful cost and latency management. Q2BSTUDIO offers precisely that combination of expertise in artificial intelligence and cloud computing to build platforms that respect the agency of all stakeholders.
Another crucial aspect is impact evaluation. Educational institutions must measure not only academic outcomes, but also agency indicators, such as how often students make autonomous decisions or their perceived control over learning. Here Business Intelligence tools (Power BI) are allies for creating dashboards that monitor these indicators and allow continuous adjustments. Likewise, automating repetitive tasks (grading exams, generating reports) can free up valuable time, but only if it does not eliminate opportunities for meaningful human interaction.
In terms of recommendations, the review suggests that educational policy frameworks should include guidelines on ethical AI use, teacher training, and ensuring equitable access. From a software development perspective, this translates into building modular and flexible platforms, where each AI feature can be activated or deactivated according to context. For example, an AI-based tutoring module can have a switch allowing the teacher to decide when and how it is used. This level of control is possible thanks to microservices architectures and well-documented API integration.
Finally, it is important to note that technology is only an enabler. Agency is not guaranteed simply by adding a virtual assistant; it requires intentional pedagogical and technical design. Software companies, like Q2BSTUDIO, have the opportunity to lead this change by offering solutions that prioritize transparency, security, and personalization. In a market where generative AI advances rapidly, those who manage to balance innovation and responsibility will be better positioned to positively impact the future of education.
